According to the existing search methods, generally image search is conducted according to keywords entered by users and/or images. In the event that the users know nothing about the description of the expected search results, the keywords entered by the users cannot accurately reflect the users' search requirements in most cases. According to another method, servers need to resolve keywords entered by the users for describing the users' search requirements, and then conduct further searches.
The existing search methods cannot directly and accurately convey the users' search requirements to the servers, and the servers have to perform numerous steps of processing the search information entered by the users. Therefore, the existing search methods have a problem of low search efficiency.